Investigating initial patterns of progression on first-line treatment in patients with de novo metastatic breast cancer.

Introduction
Although breast cancer mortality rates have declined by 43% from 1989–2020, it remains the second leading cause of cancer-related death in women [1, 2]. The majority of patients with metastatic breast cancer develop metastases months to years after an initial localized diagnosis of breast cancer. This form of disease is known as recurrent metastatic breast cancer (rMBC) [3, 4]. In contrast, 6–10% of patients with metastatic disease present with de novo metastatic breast cancer (dnMBC), classified as a primary breast tumor that is already metastatic at time of initial diagnosis [2, 3]. The majority of prior research has studied the clinical course and treatment of rMBC, with limited published data regarding unique aspects to consider when diagnosing and treating dnMBC. Patients with dnMBC are a particularly important population to study, as the relative prevalence of dnMBC continues to rise, despite the introduction of regular screening mammograms. This trend is likely largely due to the effectiveness of therapy for stage I-III disease, lowering the risk of rMBC [5, 6]. Overall, existing data suggests that patients with dnMBC have a better prognosis than those with rMBC, possibly due to biologic differences and treatment approaches [3–5, 7–9].
Small retrospective and unselected randomized studies suggest that metastasis-directed therapy (MDT) may improve outcomes in oligometastatic disease [10, 11]. However, large prospective data in breast cancer remain mixed. The NRG-BR002 trial, which included mainly recurrent MBC, showed no survival benefit with MDT [12]. Likewise, the EA2108 trial found that locoregional therapy to the intact primary in dnMBC did not improve survival and worsened quality of life [13]. More recent studies have examined MDT in selected scenarios. CURB found no progression-free survival (PFS) benefit with stereotactic ablative radiotherapy (SABR) in oligoprogressive breast cancer, though clinical subtypes varied and systemic therapy was not standardized [10]. The EXTEND trial, including patients with unselected oliogmetastatic breast cancer, also found no survival benefit. In contrast, the AVATAR trial focused on oligoprogressive HR +/HER2 − patients on endocrine therapy plus CDK4/6 inhibitors and showed a median modified PFS of 9.9 months with SABR. This supported the idea that ablating progressing sites while continuing effective systemic therapy may extend benefit in select subgroups [15]. No prospective study has evaluated whether dnMBC patients who progress only at pre-existing sites benefit from this approach including MDT. Characterization of patterns of first progression in dnMBC with identification of subgroups who may be better selected candidates for such strategies is needed.
It has been previously shown that dnMBC patients do significantly better on first-line therapy and have longer overall survival [3–5]. Thus, consideration of mechanisms to control sites of progression may have augmented impact, and time on first-line therapy might in fact be extended by MDT [16]. We hypothesize that patients with dnMBC may have distinct characteristics from the general MBC population and that some subgroups may benefit from aggressive local therapy to both the primary and metastatic sites. To date, anatomic site of first treatment failure has not been studied in patients with dnMBC. This study aims to investigate the site of first progression in patients with dnMBC and whether progression patterns differ by clinical subtype. Our goal is to inform future studies of MDT selection in the dnMBC population.

Methods

Patient population
The University of North Carolina (UNC) Metastatic Breast Cancer Database was utilized to obtain demographic and clinical data for this study. Since June 2011, all patients with metastatic breast cancer who are evaluated at UNC are entered into this prospective observational registry and followed for treatment and clinical outcomes. Participants included in this study had a confirmed diagnosis of de novo MBC, defined as stage IV metastatic disease (according to American Joint Committee on Cancer TNM staging criteria) within the first three months after the initial breast cancer diagnosis. For patients who received treatment for de novo metastatic disease prior to evaluation at UNC, data on prior therapy were collected retrospectively with subsequent clinical course followed prospectively. Patients were treated according to physician’s choice and were monitored as per standard of care, which includes regular restaging scans to assess for disease progression. We excluded patients without documented follow-up after initiation of systemic therapy. We excluded patients without documented follow-up after initiation of systemic therapy. After excluding 3 patients without documented follow-up and 2 patients without accessible records (not uploaded into the EMR), our final sample size of patients with dnMBC from the UNC Metastatic Breast Cancer Database was N = 326 with initial breast cancer diagnosis between 2011 and 2022.

Data analysis and interpretation
Baseline patient characteristics, including sex, race, and ethnicity, were analyzed utilizing descriptive statistics. Further, tumor characteristics were obtained, including ER/PR/HER2 receptor status and sites of metastases. Details regarding modality of first-line systemic therapy (endocrine vs chemotherapy vs immunotherapy vs combination) were also collected. Subsequent imaging and notes were reviewed in the electronic medical record, during which data was collected regarding the date and site of first progression by the investigators. Data for patients who received treatment outside of the UNC system was obtained from provider-uploaded medical records, including electronic health record data linkages to other health systems as well as scanned outside records uploaded into patient charts.
Location of metastasis at diagnosis was recorded and classified based on whether it was amenable for radiation treatment. Sites amenable to radiation treatment were defined as those able to be localized and targeted (i.e. brain, bone). Sites deemed not amenable to radiation treatment included those with diffuse involvement and/or lesions that could not be targeted using localized approaches (i.e. ascites, pleural lesions). Anatomic site of first progression was then coded based on site of first progression (local in-breast, pre-existing metastatic site, new metastatic site, or various combinations of the above) and described using descriptive statistics to determine the location of first progression.
Kaplan Meier survival analyses were completed to assess progression-free survival for the cohort as a whole and then categorized by clinical subtype. Cumulative incidence functions were utilized to assess time to first treatment failure for the cohort as a whole and categorized by site of first progression (defined as (a) local in-breast and/or pre-existing metastatic site vs (b) all other progression groups). These groupings enabled an analysis of progression patterns based on whether the locations involved pre-existing sites only or involved a new site. Cumulative incidence functions were then repeated to assess time to first treatment failure by clinical subtype.

Results

Patient and tumor characteristics
Table 1 illustrates the demographic characteristics of the cohort. The majority of patients were female with 322 females (98.8%) and 4 males (1.2%). Race data included 224 White (68.7%), 87 Black (26.7%), 8 Other (2.5%), 5 Asian (1.5%), and 1 American Indian or Alaskan Native (0.3%). The median age at initial dnMBC diagnosis was 56.4 years (range, 26.8–94.1). The median follow-up for the study cohort was 26.6 months (range, 0.9–297.7).
Baseline tumor characteristics and first-line therapies are listed in Table 2. The majority of tumors were of HR +/HER2− clinical subtype with 161 HR +/HER2− (49.4%), 66 HR−/HER2− (20.2%), 44 HR +/HER2 + (13.5%), and 33 HR−/HER2 + (10.1%). The first line therapy for the patients varied widely and included most commonly chemotherapy alone (30.0%), endocrine + targeted therapy (22.4%), endocrine therapy alone (18.4%), and chemotherapy + HER2 directed drugs (13.2%).

Metastatic sites present at diagnosis
Table 3 displays the metastatic sites present at diagnosis in the whole cohort. The most common metastatic sites present at diagnosis were bone (20.9%), non-local lymph node (15.0%), liver (14.8%), lung (10.9%), and brain (10.0%). Figure 1 displays the number of metastatic sites present at diagnosis per patient. Most patients (70.0%) had 2 to 5 metastatic sites, with an overall range of 1 to 11. Of sites present at diagnosis, approximately two-thirds (range 40–83%) were deemed amenable to local treatment with radiation therapy based on location.

Progression-free survival
Figure 2 displays progression-free survival (PFS) for the full cohort. Among the full cohort, PFS at 2 years was 32.7% and at 5 years, 7.8%.
Figure 3 displays PFS by clinical subtype. Among the HR +/HER2− subtype, PFS at 2 and 5 years was 37.5% (95% CI [29.6, 45.4]) and 10% (95% CI [4.7, 15.2]), respectively. Among the HR +/HER2 + subtype, PFS at 2 and 5 years was 33.8% (95% CI [18.0, 49.0]) and 8.5% (95% CI [0.0, 17.6]), respectively. Among the HR−/HER2− subtype, PFS at 2 and 5 years was 13.8% (95% CI [5.2, 22.3]) and 3.0% (95% CI [0.0, 8.5]), respectively. Lastly, among the HR−/HER2 + subtype, PFS at 2 and 5 years was 30.2% (95% CI [13.7, 46.6]) and 5.2% (95% CI [0.0, 14.7]), respectively.
Brain metastases were present in 38.7% of the cohort (n = 126) at diagnosis (Table 3). Given this distinct clinical scenario, progression-free survival analyses were stratified by the presence of brain metastases at diagnosis. For patients with brain metastases at diagnosis, PFS at 2 and 5 years was 59% (95% CI [49.6, 68.5]) and 1% (95% CI [0.0, 2.8]), respectively vs 54.9% (95% CI [48.2, 61.7]) and 14.1% (95% CI [8.8, 19.4]) for those without, p = 0.04. 64.4% of patients (n = 210) had < 5 metastatic sites present at diagnosis, while 35.6% (n = 116) of patients had 5 or more metastatic sites present at diagnosis. For patients with < 5 metastatic sites present at diagnosis, PFS at 2 and 5 years was 59.8% (95% CI [52.9,66.6]) and 11.7% (95% CI [6.8, 16.7]) respectively. For patients with 5 or greater metastatic sites present at diagnosis, PFS at 2 and 5 years was 50.3% (95% CI [40.9, 59.6]) and 4.5% (95% CI [2.0, 8.8]) respectively, p = 0.13.
Lastly, PFS did not vary significantly by time period of treatment: PFS at 2 and 5 years was 53.6% (95% CI [47.0, 60.0] and 0.9% (95% CI [0.00, 2.1], respectively for those diagnosed between 2011–2019 vs 60.0% (95% CI [42.5, 77.5] and 0.6% (95% CI [0.0, 0.9], respectively for those diagnosed between 2020–2022, p = 0.11.

Location of first progression by clinical subtype
Figure 4 displays the location of first progression of the full cohort as well as by clinical subtype. Among the full cohort, 46.5% (n = 151) initially progressed at a new site, while 40.8% (n = 133) first progressed at a pre-existing site only (local in-breast or pre-existing metastatic site or combination), and 12.6% (n = 41) had not experienced progression at time of last follow-up. A relatively similar pattern was seen throughout clinical subtype groups: among those with HR−/HER2− subtype, 36.4% (n = 24) progressed first at a pre-existing site only; among those with HR−/HER2 + subtype, 39.4% (n = 13) progressed at a pre-existing site only; and among those with HR +/HER2− subtype, 42.9% (n = 67) progressed at a pre-existing site only. A somewhat different pattern was visualized in the HR +/HER2 + individuals, in which slightly less patients initially progressed at a pre-existing site (32.6%, n = 14), with most (48.8%, n = 21) progressing at a location involving a new site.
Nine patients (3%) received local therapy with breast radiation prior to the time of first progression. Four of these patients (44%) progressed at a pre-existing metastatic site. One patient (11%) progressed at the radiated site (in-breast). One patient (11%) progressed at both the radiated in-breast site and a pre-existing metastatic site. The remaining 3 patients (33.3%) progressed elsewhere including a new metastatic site.
Among the 13% of patients (n = 41) who did not experience disease progression at last follow-up, the median age was 62.0 years (range 29.8–82.8). This subgroup was predominantly HR +/HER2− (54.8%), with smaller proportions of HR−/HER2− (11.9%), HR +/HER2 + (11.9%), and HR−/HER2 + (9.5%). First-line therapies were most commonly endocrine + targeted therapy (32.5%), followed by chemotherapy alone (20%) and endocrine alone (17.5%).

Cumulative incidence of first progression by clinical subtype
As displayed in Fig. 5, the cumulative incidence of first progression at a pre-existing site (in-breast and/or pre-existing metastatic site) for the full cohort was 32% (95% CI [26.9, 37.4]) at 2 years and 42% (95% CI [37.3, 48.8]) at 5 years. Figure 6 displays the cumulative incidence of first progression at a pre-existing site only by clinical subtype. The HR−/HER2 + subtype had a cumulative incidence of progression at a pre-existing site of 39% (95% CI [25.1, 61.2]) at 2 years and 43.8% (95% CI [29.2, 65.6]) at 5 years. The HR−/HER2− subtype had a cumulative incidence of progression at a pre-existing site of 36.3% (95% CI [26.1, 50.3]) at 2 years and 39.3% (95% CI [28.5, 54.0]) at 5 years. The cumulative incidence of progression at a pre-existing site only for the HR +/HER2− clinical subtype was 29.1% (95% CI [22.6, 37.5]) at 2 years and 45.4% (95% CI [37.8, 54.5]) at 5 years. The cumulative incidence of progression at a pre-existing site only for the HR +/HER2 + clinical subtype was 31.7% (95% CI [19.8, 50.6]) at 2 years and 34.5% (95% CI [22.2, 53.6]) at 5 years.

Discussion
In this retrospective analysis of patients with dnMBC, we found that approximately 40% of patients experienced their first progression at pre-existing sites—either in the intact breast tumor, previously known metastatic lesions, or both. This pattern of failure suggests that these patients may harbor localized disease reservoirs that are insufficiently controlled by systemic therapy alone. Furthermore, among the remaining patients, the majority progressed at both pre-existing and new sites, while a smaller proportion progressed exclusively at new metastatic locations. These findings underscore the heterogeneity in progression patterns within dnMBC and prompt further consideration of whether local therapies may delay progression and/or prolong time on first-line systemic therapy for a meaningful subset of patients.
Stratification by clinical subtype revealed subtle differences in both the timing and location of progression; however, these trends should be interpreted cautiously given the relatively small number of patients in each subtype subgroup. HR−/HER2− patients experienced the fastest progression, with a median time to progression significantly shorter than other subtypes. In contrast, HR +/HER2− patients had a more prolonged time to first progression and were more likely to recur at pre-existing sites only as their site of first failures. HR +/HER2 + patients, meanwhile, more frequently progressed first at new sites, suggesting a distinct pattern of systemic dissemination. These subtype-specific trends reinforce the biologic diversity of dnMBC and may have direct implications for therapeutic decision-making, particularly in identifying candidates for early local intervention.
Our data are consistent with and expand upon previous findings regarding the phenotypic behavior of clinical subtypes. The observed late progression in HR +/HER2− patients align with prior studies indicating that this group typically follows a slower, more indolent course [17]. Conversely, the rapid progression in HR−/HER2− patients reflects their known aggressive trajectory [18]. Interestingly, our analysis revealed a contrast between the HR +/HER2− and HR +/HER2 + groups, with the former more likely to experience pre-existing site-only progression as the site of first failure, and the latter showing more new metastatic sites as site of first failure. These findings suggest that subtype not only informs systemic therapy decisions but may also predict spatial progression patterns, potentially guiding local treatment strategies.
Additionally, the subgroup of patients who did not experience disease progression (n = 41, 13% of the cohort) was made up primarily of HR +/HER2 − tumors, consistent with a more indolent disease trajectory. This subgroup predominantly received endocrine-based first-line therapy, suggesting that both tumor biology and treatment selection may contribute to prolonged disease stability.
While most research has historically centered on rMBC, dnMBC is becoming increasingly prevalent and deserves independent consideration. Although dnMBC and rMBC are often treated similarly, existing literature indicates that patients with dnMBC have better long-term outcomes, including improved overall survival and longer disease-free intervals [3–5, 7–9]. These prognostic advantages may reflect fundamental biologic or clinical differences, supporting the argument that dnMBC represents a distinct clinical entity. By focusing on patterns of first progression in dnMBC, our study contributes new insights that may help refine treatment paradigms specifically for this subgroup, including selection of patients who might be most likely to benefit from MDT.
The potential role of MDT, such as SABR, in oligometastatic disease has been increasingly studied. The SABR-COMET trial demonstrated that MDT can significantly improve progression-free and overall survival in patients with oligometastatic disease [16]. Although the trial included breast cancer patients, they were a minority, prompting more specific studies like NRG-BR002. NRG-BR002, which included both rMBC and dnMBC patients, found no survival benefit with MDT [12]. However, only 22% of enrolled patients had dnMBC, and the study may not have been adequately powered to detect differences within this subgroup. Furthermore, in the MDT arm of NRG-BR002, approximately 90% of treated lesions remained controlled throughout the study period, highlighting the effectiveness of local therapies in preventing progression at targeted sites. More recently, the AVATAR phase II trial evaluated SABR in patients with oligoprogressive HR +/HER2 − metastatic breast cancer maintained on endocrine therapy and CDK4/6 inhibitors, reporting a median modified progression-free survival of 9.9 months [15]. These findings suggest that targeting limited sites of progression can prolong the benefit of first-line therapy, a strategy that may be especially relevant in dnMBC given its often indolent disease course and reliance on durable systemic control.
Based on the metastatic sites present at diagnosis in our cohort (Table 3), roughly two-thirds of metastatic lesions present at diagnosis would have been amenable to targeted therapy with radiation. By delaying progression at pre-existing sites, MDT could potentially extend time on first-line systemic therapy, thereby preserving treatment options and potentially improving quality of life. In addition, given the distinct behavior of dnMBC, particularly the presence of an intact primary tumor and relatively high rates of pre-existing site progression (~ 40%), our findings suggest that MDT (with treatment of both the primary and amenable metastatic sites – unlike EA2108 which only addressed the primary breast tumor) may have greater utility in dnMBC than previously appreciated.
Our findings also inform an important clinical question that remains unanswered by existing trials: for dnMBC patients who experience first progression exclusively at pre-existing sites, is it more beneficial to switch systemic therapy or to continue the current regimen while using MDT to ablate the progressing lesions? Current evidence offers mixed signals. The CURB and EXTEND trials both failed to show a survival advantage for SABR in broader metastatic breast cancer populations, though neither was restricted to oligoprogression or to dnMBC, and systemic therapy strategies varied [10, 14]. In contrast, the AVATAR trial demonstrated that in oligoprogressive HR +/HER2 − disease, continuing endocrine therapy and CDK4/6 inhibitors while treating progressing lesions with SABR prolonged modified PFS, supporting the concept that MDT can extend the benefit of effective systemic regimens in select patients [15]. Given that approximately 40% of our dnMBC cohort progressed first at pre-existing sites, this subgroup may represent prime candidates for such an approach. Definitively addressing this question would require a large, randomized trial specifically designed for dnMBC, stratified by clinical subtype, and powered to assess both oncologic and quality-of-life outcomes.
Our study has limitations, including its retrospective nature and reliance on electronic health record documentation to determine the timing and location of first progression. While efforts were made to minimize bias—including consistent chart review using a standardized approach—misclassification and missing data remain possible. Additionally, our single-institution cohort may not fully capture the diversity of the broader dnMBC population. Certain analyses, including subtype-specific progression patterns, were based on small subgroup sizes and limited numbers of events, which restricts the statistical reliability of these observations and renders these findings descriptive and hypothesis-generating rather than definitive. A small subset of patients (3%) received local therapy with breast radiation prior to first progression, which may have influenced progression patterns and confounded associations between site of first progression and clinical subtype. However, the clinical and tumor characteristics of these patients were generally similar to those of the broader cohort, suggesting limited impact on overall trends. Future studies are needed to explore the impact of MDT and validate the present findings in larger, multicenter cohorts with diverse demographic and clinical characteristics.
In conclusion, our data demonstrate that a significant subset of dnMBC patients experience first progression at pre-existing disease sites. These results support further investigation into the role of MDT (to both the intact breast tumor and sites of metastases) in this population, ideally through prospective, randomized studies designed specifically for dnMBC. Such trials should evaluate not only oncologic outcomes, but also the impact of local therapies on patient-reported quality of life and long-term treatment trajectories.